Most Amazon sellers start keyword research with a tool. Type in a seed keyword, export a list, pick the ones with high volume and medium competition. Done.
That process finds keywords. It does not find the right keywords for your specific listing — the terms your actual customers are using, the queries your conversion rate already favors, the gaps where dominant competitors rank that you do not. Those insights require a different starting point entirely.
At OBG, keyword research starts with your own data. Then we go to the tools. Then we look at the competition. In that order, for a reason.
Step One: Start With SQP — What Your Customers Actually Searched
Amazon's Search Query Performance (SQP) report, available through Brand Analytics for registered brands, is the most valuable and most underused keyword research tool on the platform. It does not show you what people are searching in general. It shows you what people searched when they ended up purchasing your product specifically.
That distinction is everything. A tool like DataDive can tell you that "stainless steel water bottle 40oz" gets 85,000 searches per month. SQP tells you that 34% of your own purchases came through that exact query — and that your conversion rate on that search term is 14.2%, while the market average is 9.8%. That is actionable information. You know you are above market CVR on that term, which means you should be pushing it harder in PPC and ensuring it is prominent in your listing.
The SQP data we pull on every account covers three things we act on immediately:
- High-CVR search terms not in our exact match campaigns. If a term is driving purchases but we are only capturing it in auto or broad match, we graduate it to exact match immediately and build toward ranking for it organically.
- High-volume terms where our CVR is below market average. These are terms where the listing is not resonating — the customer's intent from that search does not match what our listing delivers. We investigate whether the title, main image, or bullets need adjustment before we push harder on those terms in PPC.
- Terms we are not capturing at all. SQP shows us what customers searched before buying our product. If there are high-volume terms driving purchases that are not in any of our campaigns, that is a gap to fill immediately.
SQP is step one of our Avatar Alignment process — the foundation for understanding who is actually buying your product, not who you think is buying it. The search terms customers use before purchasing are a direct window into their intent, their language, and their decision-making process. No other data source gives you that as cleanly.
Step Two: Layer DataDive for Volume, Competition, and Gaps
Once we have the SQP baseline — the terms that are already driving purchases — we move to DataDive for keyword expansion. DataDive gives us search volume, keyword difficulty, organic rank data, and competitor keyword coverage that we can cross-reference against our existing performance data.
The specific questions we use DataDive to answer:
What high-volume terms are we not ranking for organically?
We pull organic ranking data for the top 50–100 keywords in the category and identify terms in the top quartile of search volume where our listing is not on page one organically. These become research campaign targets — broad or phrase match campaigns with limited budgets designed to test whether we can convert for those terms before committing to ranking campaigns.
Where does our listing have ranking leverage we have not activated?
Sometimes a listing is ranking organically on page two or three for high-volume terms — just outside the visibility window where most clicks happen. A targeted PPC push on those specific terms, combined with listing optimization, can tip the listing onto page one organically. That is a high-ROI keyword investment: a small amount of ad spend to catalyze organic rank that then self-sustains.
What are the long-tail opportunities with low competition and solid volume?
The 50,000-search-per-month term gets all the attention. The ten 3,000-search-per-month terms with lower competition often drive a higher percentage of revenue collectively — and are far easier to rank for. DataDive's keyword clustering shows us where the long-tail volume concentrates so we can build campaigns around it systematically rather than cherry-picking obvious head terms.
Step Three: Look at Who Is Already Winning — and Why
Your best keyword strategy may already exist. You just need to look at the competitors who are ranking for terms you want and understand exactly how they got there.
Instead of asking "what should we do?", we ask "who is already winning — and why?" Competitors leave clues: which keywords their title prioritizes, which angles their main image emphasizes, which benefit their A+ content leads with. You do not need to copy. You need to understand what the algorithm and the customer are responding to.
With DataDive and Datarova, we run a competitor keyword coverage audit on every account: pull the top 5 organic ranking competitors in the category, map every keyword they rank for in positions 1–20, then cross-reference against our own ranking data. The gap — keywords they rank for that we do not — becomes our strategic opportunity list.
Prioritization within that gap comes from three filters:
- Search volume — the terms worth fighting for, not just the ones that are easy.
- Our own CVR data from SQP — terms where our conversion rate suggests we can compete if we get the visibility.
- Competitor listing alignment — terms where the top-ranking competitor's listing is weak or their reviews indicate customer frustration we do not have. Those are the opportunities where outranking them is achievable.
Step Four: Feed Keywords Back Into the Avatar Alignment Framework
Keyword research at OBG is not a standalone exercise. It feeds directly into our Avatar Alignment process — the framework we use to ensure your listing is built around the language, intent, and buying triggers of your actual customer.
Here is how it connects: SQP data tells us how customers describe their need before purchasing. DataDive tells us the volume and competition for those descriptions. Competitor analysis tells us what positioning is already winning in the category. That combined picture forms the basis of the customer avatar — who they are, what they want, how they talk about it, and what they need to see in a listing to convert.
From that avatar, we create three listing variants emphasizing different angles — different benefit hierarchies, different headline keywords, different proof points in the bullets. We split test through Jungle Ace. The winner becomes the canonical listing. The headline image uses the highest-converting keyword from the SQP data as the primary visual anchor.
Keywords are not just advertising targets. They are insight into your customer's mind. The brands that treat them as both — advertising assets and customer intelligence — build listings that convert for the right reasons, not just listings that rank.
Common Keyword Research Mistakes That Cost Rankings
After auditing hundreds of accounts, the same errors come up repeatedly:
- Targeting only high-volume head terms. Head terms are the most competitive, the most expensive to rank for, and often the least specific — meaning the customer using them may be earlier in the purchase decision than you want. Long-tail terms with clear purchase intent often drive better CVR at lower competition.
- Ignoring backend keyword fields. Amazon's backend search terms, subject matter fields, and intended use fields are indexed by the algorithm but do not appear in your listing copy. Many brands leave these empty or fill them with generic terms. We fill every available character with relevant, non-duplicate terms from our keyword research, prioritizing terms that do not fit naturally into the listing copy.
- Never revisiting the keyword strategy. Customer language evolves. Category terminology changes. New competitors enter and establish new keyword associations. A keyword strategy built 18 months ago and never revisited is working with stale data. We run a full keyword audit with updated SQP and DataDive data every quarter minimum.
- Treating all keywords equally. Not all keywords deserve the same investment. Terms where your CVR is above market average deserve aggressive PPC spend and active organic ranking pushes. Terms where your CVR is below market deserve investigation before investment. The keyword strategy should be weighted by performance data, not by volume alone.
What Keyword Research Looks Like at the Start of an OBG Engagement
When we take on a new account, the keyword research process runs in the first two weeks before we restructure any campaigns. We pull SQP data for the trailing 90 days, run a DataDive keyword audit, execute a competitor ranking gap analysis, and map the results against the existing campaign structure.
Consistently, we find three things: keywords that should be in exact match campaigns that are only covered by auto or broad match, terms with above-market CVR that are not receiving adequate budget, and significant spend on terms where the listing is not converting well — money that would be better redirected to proven performers.
Streetwise Security came to us with a keyword strategy built on generic category terms with no SQP grounding. We rebuilt the keyword architecture from the SQP data up, and the account saw 50% growth in both sales and profit year over year. The budget barely changed. The targeting did.
Work With OBG
Keyword research is not a task you complete once. It is the ongoing intelligence layer that keeps your PPC strategy aligned with how real customers are actually shopping in your category right now. We do this work continuously — SQP review, competitor gap analysis, DataDive keyword audits — as part of every managed account.
OBG offers a free strategy call and a 30-day profitability guarantee. If we cannot show you a clear path to better keyword coverage and stronger organic rank in the first 30 days, you do not owe us for that time. Schedule your free strategy call and let us show you what is in your own SQP data that you are currently leaving on the table.